use comma separated arff instead of bags of features

workflows/ilp/aleph/aleph.py

@@ -217,10 +217,11 @@ class Aleph(object):
         cat('@DATA')
         examples = re.findall(r"example\((\w+),(\[[\d,]*\]),(\w+)\)\.", pl_dataset)
         for _, features, cls in examples:
-            vals = ['%d 1' % (int(fid)-1) for fid in json.loads(features)] + \
-                   ['%d %s' % (class_id, cls)]
-            cat('{%s}' % ', '.join(vals))
-
+            vals = []
+            for i in range(1, class_id):
+                vals.append('1' if i in json.loads(features) else '0')
+	    vals.append(cls)
+            cat('%s' % ','.join(vals))
         return arff.getvalue()